Optimizing COCOMO II parameters using artificial bee colony method
Rayandra Yala Pratama, Riyanarto Sarno, Sholiq Sholiq · 2017
Cost estimation is a crucial and essential process in software industry. The more accurate cost estimated, the more efficient the project became. This cost estimation become a challenge for software industry to bring accurate result. There are many methods to solve this problem. Constructive Cost Model is usual method that is used to estimate software cost. This model was proposed in 1981 by using regression analysis with 63 types of project data. In 2000, COCOMO II was introduced. This new model of COCOMO use cost drivers, scale factors, and project size that measured by line of code. COCOMO II has 4 parameters A, B, C and D. However, using this parameters are not guarantee accurate result. This paper proposed Bee Colony Optimization to calibrate the COCOMO II model parameter to be more accurate for effort estimation. This Bee Colony Optimization is applied on Nasa93 dataset that consisted of 93 projects which each project has 22 cost drivers, project's size, effort, and development time. This proposed method gives MMRE result 50.584% on effort and 14.192% on development time.